SlimeNENC — Legacy modernization
BLACKBOX — bring it on.
COBOL modernization does not need a “understand the business logic” phase.
Treat the source itself as the specification, and transpile bit-exactly under a deterministic rule base. That’s all.
Discard meaning extraction
The industry standard play — have analysts reverse-engineer 30 years of implicit rules — collapses at scale, as CBA, IRS, and TSB Bank’s hundred-million-dollar failures show. We treat the source itself as the spec and move the bytes through unchanged.
We do not violate “No Silver Bullet”
Brooks (1986) said complexity cannot be removed — he did not say “therefore humans must re-express it.” Our approach belongs to the same lineage as virtualization, containers, and WASM: preserve and transport, without re-expression.
Algebra, not probability
Competing LLM-based migration tools (GitHub Copilot for COBOL / IBM watsonx / AWS Q) work by probabilistic inference, so the same input produces a different output every run. We capture meaning in a mathematical fixed structure, guaranteeing bit-for-bit identical output every time.
Tricky behaviours automatically detected
Timing-sensitive operations (waits, cursors, channel control) are auto-detected and audit-marked. We also ship reverse-engineering tools that recover C source from binaries for customers who have lost the source (ELF 25/25 + PE 21/21 byte-exact validated).
Measured (updated 2026-05-20 PM):
The US federal COBOL validation suite (FIPS 21-3) plus 11 external corpora — including the 1976-born, ~100K-line ACAS accounting system — transpiled into 5 target languages, 4,595 / 4,595 match bit-identically (100.000%). Anyone can reproduce the result inside a WASI sandbox.
Additionally, US federal Medicare pricing across 7 systems, 675 / 675 (100.000%) auto-converted (output-level match against the official Java is being verified next).
A rare implementation that can honestly say “100% is a measured count, not a probability.”
SlimeCOBOL product → Japanese mainframe PoC → Legacy conversion service →
Even from source-lost COBOL binaries — we ship both COBOL and Java
An estimated 12–24 billion lines of COBOL belong to customers who have lost the original source. LLMs cannot infer meaning from binaries; Accenture and Capgemini decline the work as “unestimable” — a monopoly market. Our new SlimeRESCUE-VOS3 reads the executable directly and emits Java/COBOL/Rust, cross-validated 6/6 byte-exact against the IBM-recognised emulator. Hitachi VOS3 / Fujitsu MSP / IBM z/OS share one code path.
1/3.5 the size, harder to reverse-engineer — every conversion still bit-identical
We re-implemented the 8 core modules that turn COBOL into Java in Rust, a sturdier language. Total binary size shrank from 2.16 MB to 637 KB. We tested 1,818 COBOL files across 5 target languages — every single result came out bit-identical (100.000%). Rust’s built-in safety also makes reverse engineering noticeably harder.
4K, at 1.79 Mbps.
The eye cannot tell.
Slime-Boy: let me explain what that number means — gently ▾
“4K at 1.79 Mbps” — what does that mean?
- 4K = ultra-sharp video filling your TV screen (3840 × 2160 pixels, four times Full HD)
- Mbps = how much data flows per second. Bigger number = thicker water pipe
- Normal 4K video needs a thick pipe (25 – 50 Mbps) — Wi-Fi or fiber territory
- But JAVATEL's NormAV1 flows 4K through a tiny 1.79 Mbps pipe!
- That's roughly 1/14 to 1/28 of normal — one LTE 4G line is enough
How? Math (the commutator norm) finds where your eye is paying attention — and where it isn't — then sends bits only where they matter. We'll explain more further down the page!
Live switch demo → 30-day trial → Latest news →
Patents pending: JP 2026-046898 / 046609 / 046620 / 046625 · Paper: Commutator Norm as a Continuous Criterion for GOP Safety and Perceptual Quality Redistribution in Video Encoding — IEEE TCSVT v8 submission
Four product pillars — led by shortest market reach
PSDP — Same-Language Parallelization
For finance, regulated industries, cloud cost reduction, HFT, scientific computing, ML inference. 378 consecutive runs all bit-identical; any mismatch triggers an instant rollback.
Slime-Boy: 4× speedup without changing my source? I don't believe it... ▾
“No source rewrites” — how is that possible?
- Your
OrderBatch.javastays byte-for-byte identical (git diffshows nothing) - At build time, PSDP reads the bytecode and injects a parallel version alongside (Java agent / .NET profiler / Rust proc-macro)
- 4 cores → parallel path. 1 core → sequential. No dispatch code on your side
- PSDP’s proprietary mathematical synchronization mechanism excludes the “parallel changes my answer” failure mode by theory (details in patent filings; redacted here)
- Every production run, bench compares sha256 of sequential vs parallel; mismatch → instant rollback
Result: not one line of your code touched / just rebuild / output bit-identical / 4× on 4 cores (measured 2.97–4.23×). For audit, “zero code change” is the selling point. PSDP overview →
PSDP overview → SlimeNENC customers: PSDP tool deployment is free; PSDP runtime is 2%/year of conversion cost.
NormH.264 / NormAV1 — 4K to 1/4
Math finds where the eye is looking and concentrates bits there, taking them from regions that nobody is looking at. NormH.264 is production; NormAV1 is experimental.
Legacy → Java bit-exact
COBOL / JCL / MUMPS / PL/I / RPG / FORTRAN / Natural / assembler / Linux ELF / Windows PE — 10 source types transpiled into Java. US federal validation 501 / 501 pass, with a tamper-detection audit chain. See the price table.
Sub-second WebRTC + 48h time-shift
JAVATEL OEM of Unreal Media Streaming. Plugin-free browser playback at under 1 second of latency, with 48 hours of rewind.
Evidence — measurements
| H.264 plain | 25 – 50 Mbps |
| AV1 plain (AOM) | 12 – 15 Mbps |
| NormH.264 | 6 – 8 Mbps |
| NormAV1 | 1.79 Mbps |
Measured on 4K 60fps urban footage. VMAF 84.36 (industry-standard quality metric; higher is better).
Verified on iPhone, Intel laptop, and AMD desktop — all play 4K at 30 fps smoothly.
- Math finds where the eye is looking — and sends bits only there
- Plugs straight into existing H.264 / AV1 — no new player required
- Even already-compressed video shrinks another 8.43×
NormAV1 sweep demo · dav1d WASM live demo · 2-core verified on real devices
Latest news
All news →- SlimeNENC engine rewritten in Rust — 1/3.5 the size, harder to reverse-engineer, every conversion still bit-identical.
- Recovering COBOL and Java from the binaries of mainframes whose source is lost — new SlimeRESCUE-VOS3, 6/6 byte-exact against the IBM-recognised emulator.
- US federal Medicare pricing programs auto-converted into 5 languages — a one-cent miscalculation is a federal-law violation: 675/675 success.
- SlimeCOBOL reorganised as a 5-SKU umbrella — SlimeCOBOL-MF / -GnuCOBOL released today.
- A 1976-born ~100K-line accounting system: fully converted — 685/685 = 100% match.
- US Federal COBOL Validation (FIPS 21-3) cleared — 10 corpora, 782 files transpiled into 5 languages, 3,910 / 3,910 match.
- SIer / Reseller Partner Programme launched — online application open at /partners/register/.
- PSDP 4-way pricing finalised — see the price table.
- PSDP released as an independent product line — same-language parallelization, separated from SlimeNENC family.
- SlimeASM released — z/Architecture HLASM + x64 MASM → Java bit-exact.
- NormH.264 v0.4.2 released — Trial Signature Mode bundled; AV1 moved to NormAV1.
- SlimeNatural released — Software AG Natural / ADABAS → Java + PostgreSQL bit-exact.
- SlimeFORTRAN 9-language multi-target completed.
JAVATEL Inc. has delivered surveillance camera systems, video management, and media streaming across Japan since 1997. Headquartered in Nishinomiya, Hyōgo. The Video Intelligence in our company logo is not a recent trend — it is the direction we have walked for more than 29 years.
Partners: Genetec (longest-tenured authorized partner in Japan) / XEOMA AI NVR / Planet Networks (Taiwan) ·
Why Javatel · Partners · Contact